AI Drives Surge of Self-Represented Lawsuits

AI-fueled filings surged to 16.8 percent of federal cases in 2025, yet self-represented litigants still don't win more often.

A new study analyzed 4.5 million federal civil cases from 2005 to 2026 and found that the share of lawsuits filed by people without lawyers rose from 11 percent in 2022 to 16.8 percent in 2025, with filings by non-lawyers more than doubling since before 2023. The team reports that the jump appears tied to increasing use of AI in the legal workflow, not just a spike in individual behavior. The paper shows that while AI seems to expand access to filing, it does not translate into higher win rates for the unrepresented. Benchmarks indicate the uptick in self-representation tracks with broader AI adoption in the judiciary and related services.

In Colorado, federal magistrate Judge Maritza Braswell describes a courtroom already under strain from more filings, many from people who have no attorney. She says she uses AI to vet court documents and has learned to recognize patterns and quirks in AI produced text. “I correlate that to AI in part because I see AI use,” she explains, noting that AI has helped with drafting quality in some pleadings but also that she has encountered hallucinated cases and fabricated quotes. It is a striking example of how technology can both lower barriers to entry and complicate the judge’s task of separating plausible claims from fiction. The team reports that such observations are not isolated to one court; they reflect a broader swing toward AI-assisted drafting and review across the system.

The broader implication is a mix of accessibility and risk. The paper shows that the same AI tools that help a claimant file a complaint without a lawyer can also propagate errors, misstate facts, or rely on dubious authorities. As courts grapple with a deluge of self-represented cases, lawmakers and judges are asking who bears responsibility when AI provides bad legal guidance. Across the U.S., a growing chorus of policymakers is examining accountability for AI-provided legal advice and who should foot the bill when missteps occur.

From a practitioner’s lens, the wave creates tangible constraints and tradeoffs. First, triage becomes a core engineering problem: how to automatically flag potentially reckless or misleading AI-generated pleadings before they clog the docket, without stifling legitimate access. Second, the quality versus speed tradeoff matters: AI can speed up initial drafting and formatting, but it can also introduce invented quotes, erroneous citations, or mischaracterizations of precedents if not carefully monitored. Third, failure modes are real and repeatable: hallucinated quotes and false court references are not edge cases in this data set, but a recurring threat that requires human verification. Fourth, policy watch next: without clear standards for AI-authored content, courts risk uneven outcomes and escalating litigation costs, while legislators weigh who should bear the financial and ethical responsibility for bot-dispensed legal advice.

The surge is not doom or triumph in isolation; it is a signal that the courtroom ecosystem must evolve. If AI remains a tool for access rather than a shortcut to justice, these design choices, such as verification workflows and funding for judicial AI literacy, will define whether the trend strengthens or weakens the rule of law.